A multichannel, semiautomated, blood cell counting system (Coulter Counter Model S550) was modified for use in veterinary hematology by increasing both the erythrocyte and leukocyte aperture currents to 225 V and 195 V, respectively, followed by calibration with human blood. It was evaluated by use of 350 samples from dogs, cats, horses, and cows. Values for leukocyte count, erythrocyte count, mean corpuscular volume, and hematocrit generated by the S550 were compared with values generated by an automated multichannel counter with histogram capability and other reference procedures when appropriate. Mean differences for values between S550 and reference values were less than calibration tolerance limits for the instrument. Correlation coefficients were excellent for all values of each species. To assess behavior of leukocytes of the different species with respect to the counting threshold, leukocyte size distribution histograms were generated for all samples analyzed on the S550. Means for mean leukocyte volumes in diluent and lysing reagents were 55.5, 56.6, 67.4, and 72.8 fl for dogs, cats, horses, and cows, respectively. Canine leukocyte counts, because of small leukocyte size, were an average of 14% less for 5 samples analyzed on the unmodified instrument, compared with analysis after increasing the leukocyte aperture current. Leukocyte threshold failures attributable to interfering particles, resulting in falsely high counts, were recognized in 14%, 10%, 8% and 0% of feline, bovine, canine, and equine samples, respectively. The magnitude of error in these samples averaged 5% for cows and dogs, but was considered not important. However, leukocyte counts of feline samples in this group averaged 44% falsely high.